卷积神经网络用于区分良性和恶性瘤与多中心国际计算机断层扫描数据集
Michail E Klontzas1,2,3, Georgios Kalarakis4,5, Emmanouil Koltsakis4
1Department of Medical Imaging, University Hospital of Heraklion, Heraklion, Crete, Greece.
Insights into imaging
|January 25, 2024
概括
卷积神经网络 (CNN) 可以使用CT扫描准确区分良性和恶性瘤. 开始-ResNetV2模型显示了最高的性能,有助于精确的瘤分类.
科学领域:
- 放射学和医学成像学 医学成像学
- 人工智能在医学中的应用
- 在瘤学瘤学.
背景情况:
- 在CT扫描上区分良性和恶性瘤是诊断上具有挑战性的.
- 准确的分类对于适当的患者管理和治疗计划至关重要.
研究的目的:
- 评估卷积神经网络 (CNN) 的有效性,以区分良性和恶性瘤.
- 利用多样化的多机构,多供应商和多中心对比增强CT数据集进行培训和验证.
主要方法:
- 三个CNN架构 (InceptionV3,Inception-ResNetV2,VGG-16) 在264个经组织学确认的瘤中使用转移学习进行了微调.
- 使用了数据增强和70%:30%的火车测试分割. 使用从接收器运行特征 (ROC) 曲线的曲线下的面积 (AUC) 来评估性能.
- 突出地图被生成以可视化表现最好的CNN的决策过程.
主要成果:
- Inception-ResNetV2获得了0.918的最高AUC,在区分瘤类型方面表现出卓越的表现.
- 发明V3和整体模型的AUC为0.894,而VGG-16的AUC为0.813.
- Saliency 地图表明,Inception-ResNetV2 专注于瘤特征和瘤 - 帕伦基马接口.
结论:
- 深度学习模型,特别是Inception-ResNetV2,可以准确地区分良性瘤和恶性瘤.
- 在多样化,多中心数据集上培训CNN增强了他们概括和有效执行的能力.
- 在临床实践中,CNN为提高瘤分类的准确性提供了一个有希望的工具.
相关概念视频
Imaging Studies I: Kidney, Ureter, and Bladder Studies
Kidney, Ureter, and Bladder (KUB) StudiesKidney, Ureter, and Bladder (KUB) studies are standard diagnostic imaging procedures used to assess the anatomy of the urinary system. They are commonly utilized for patients experiencing abdominal pain or urinary symptoms. By using a simple X-ray of the abdomen, KUB studies can reveal structural and pathological abnormalities within the kidneys, ureters, and bladder. These studies are particularly valuable in diagnosing kidney stones, urinary...
Imaging Studies III: Computed Tomography
DefinitionComputed Tomography (CT) of the genitourinary (GU) tract is a non-invasive imaging modality that utilizes X-rays and computer processing to generate detailed cross-sectional images of the urinary system, encompassing the kidneys, ureters, bladder, and adjacent structures such as the adrenal glands.PurposeCT scans of the GU tract serve several diagnostic and therapeutic purposes, including:Diagnosis of Urinary Tract Diseases: Detects kidney stones, tumors, cysts, and congenital...


